# Narratio v1.5.0

Narratio v1.5.0 makes repeated post-transcript artifact development faster and
more explicit while retaining the fixed, stage-driven pipeline model.

## Highlights

- The canonical pipeline now completes deterministic rendering before
  extraction, cleanly separating transcript-generating stages from
  artifact-generating stages.
- `narratio run` and `narratio session plan` accept inclusive `--from` and
  `--through` bounds. Excluded transcript stages are not executed or
  invalidated by a bounded artifact-regeneration run.
- `narratio regenerate-artifacts SESSION` is an exact convenience alias for a
  forced run from `extract` through `analyze`, including focused
  `--artifacts` selections.
- Configured Scriptorium artifacts now have independent,
  manifest-authoritative freshness. Narratio reuses validated current work,
  rebuilds stale prerequisites in dependency order, and persists successful,
  failed, and newly stale artifact state when an analysis invocation only
  partially succeeds.
- Publish consumes only configured artifacts backed by current manifest
  evidence; incidental or tampered files are not promoted as current output.

## Reliability And Administration

- Bounded prerequisites are checked again under the session lock before any
  run mutation, closing a concurrent-run race.
- Analysis fingerprints are stable across executable and configuration path
  changes and continue to cover only Narratio-observable semantic inputs.
- Runner composition now carries one validated execution plan from command
  parsing through prerequisite validation, adapter composition, manifest
  recording, and stage execution.
- `narratio version` reports the exact tag embedded in official release
  binaries; ordinary source builds report `dev`.

## Upgrade Notes

- Existing unbounded commands and direct `run-stage`, `analyze`, and `publish`
  workflows retain their meanings.
- Manifests written before artifact-level analysis state remain readable.
  Legacy aggregate analysis success is not sufficient freshness evidence, so
  the first analysis evaluation after upgrading may regenerate configured
  artifacts once.
- Narratio cannot observe executable contents or configuration, prompt,
  profile, module, and other files loaded privately by Scriptorium. Explicitly
  force affected artifacts after changing those private inputs.